Code agents

Production-grade, multi-tenant chat service with FastAPI + WebSockets, OpenAI/Anthropic orchestration, semantic caching, and K8s/IaC deployment, Includes observability (Prometheus/Grafana/Jaeger), FinOps cost tracking, and DR runbooks.

Fix text instantly with AI — one hotkey, perfect writing. Supports OpenAI, Anthropic, Ollama, AWS Bedrock.

Tiered memory for AI — episodes, facts, preferences, reflections. Local. Portable. Yours.

Local AI Proxy Management Plugin for JetBrains Rider IDE - Configure multiple AI models, control which models are available, and manage API keys securely

Privacy-first Claude Code burn-rate analyzer. Predict weekly limit hits, spot model-swap savings.

🤝 Elevate collaboration with Claude Cowork, an open-source tool that enhances productivity through advanced AI capabilities for coding and digital tasks.

Enhance the LinkedIn user experience with AI

Bridge Claude Code CLI to a REST API. Leverage your Claude Pro/Max subscription as a free API with FastAPI, featuring a built-in Web UI and multi-directory support.

🛠️ Enhance programming workflows with a curated library of agent skills for tasks like code review, debugging, and planning for efficient project execution.

Get notified when Claude Code finishes responding. Sound, notification banner, or both.

A terminal-native, multi-provider intelligent assistant that plans, executes, and tracks developer tasks, not just answers prompts, similar to Claude Code and Gemini CLI.

Elenchus MCP Server - Adversarial verification system for code review

Drop-in fix for OpenClaw — route agents through Claude Code CLI on your Max subscription. No API key, no per-token billing.

your calendar, but yapping. Apple Calendar MCP server for Claude Desktop & Claude Code — 28 voices (Mom, Future-you, Garry Tan, LeBron James, Hilary Hahn, Alan Turing, Confucius, ...) yap on every event using your past calendar memory + web research. macOS-only, local-only, fully reversible.

Spec-driven development for Claude Code. Design docs before code, structured pipeline with quality gates, zero infrastructure.

Programmable runtime for Claude Code with persistent state, context, control flow, and verification.

One command. Your entire Claude Code setup. Done.

Real-time statusline plugin for Claude Code - track GitHub PRs, control music, get smart alerts

AI-powered firewall log analysis platform. Upload Cloudflare/nginx/Apache logs, get OWASP-categorized threat reports with PDF export.

AI-powered documentation RAG system with MCP server for Claude Code. Search and retrieve technical documentation on-demand with vector embeddings and smart web scraping.

Browse, search, and resume Claude Code chat sessions from your terminal.

A Retrieval-Augmented Generation (RAG) based multilingual chatbot specialized in nutrition & diet planning. It combines powerful NLP models with a curated nutrition knowledge base to provide personalized diet recommendations, meal plans, and fitness tips based on user input (age, weight, activity level, health goals, dietary preferences).

A Retrieval-Augmented Generation (RAG) system designed to provide answers from the arabic datasets. This project is done on static news archives using RESTful API and a web UI.

Dynamic Intelligence Graph (DIG) MCP server for AI coding agents. Persistent code knowledge graph, a code RAG improvement, with hybrid Graph + Vector + BM25 retrieval, agent memory, multi-agent coordination, and impact-scoped test selection. Works with Claude Code, VS Code Copilot, and Cursor.

An intelligent chatbot that answers questions about your CV using Retrieval Augmented Generation (RAG)

IntelliSearch is an advanced retrieval-based question-answering and recommendation system that leverages embeddings and a large language model (LLM) to provide accurate and relevant information to users.

A CLI that takes natural language and converts it to bash commands using a local code generation model or a curated prompt-based rule engine.

A runtime governance, security and execution control layer for Retrieval-Augmented Generation (RAG) systems

An Agentic RAG-powered chatbot delivering intelligent, fact-grounded and context-aware answers for the DIEM Department at the University of Salerno.

A genral RAG Search chatbot, with SoTA RAG techniques such as HyDE, Hybrid retrieval with BM25 + RRF and Cross encoder reranking. Evaluated on the BEIR scifact dataset and compared all the different pipelines i tried along the way

API-difycall 是一个轻量级的 Python 项目，用于简化与 Dify API 的交互。Dify 是一个强大的 AI 应用开发平台，本库旨在让开发者能够更便捷地集成和使用 Dify 的各项功能。

SOC Analyst Automation using a RAG model integrates a knowledge retrieval system with generative AI to automate SOC Level-1 tasks. It processes server logs, retrieves relevant security insights, and generates accurate responses, enhancing incident analysis, reducing response times, and improving efficiency in handling cybersecurity threats through

An Intelligent SOC Ecosystem that integrates SIEM, SOAR, and SmartXDR to enhance monitoring, detection, and response to cyber attacks. SmartXDR combines OpenXDR tools (Wazuh, Suricata, Zeek, pfSense,…) with AI/ML to reduce false positives, prioritize critical threats, and accelerate incident response, ensuring a more adaptive and effective SOC.

A lightweight, local Retrieval-Augmented Generation (RAG) system for domain-specific Q&A over financial documents. Uses pdfplumber for PDF parsing, sentence-transformers for dense retrieval, and optionally connects to local LLMs (e.g., Ollama + Mistral). Runs on FastAPI with a custom frontend.

💬 Engage in continuous chat while dynamically building a JSONL playbook, enhancing insights and strategies with each interaction.

This Graph RAG Application is a web-based tool that allows users to ask questions about the Mission Impossible film franchise and receive detailed, contextually relevant answers. By combining retrieval-based methods with generative AI, the application ensures that responses are both accurate and engaging.

Code for the presentation.

The AI-Based Pre-Screening Assistant Bot is an AI-powered technical interview assistant that evaluates candidates using adaptive questioning, real-time feedback, and automated report generation. Built with Streamlit and RAGPipeline, it dynamically adjusts question complexity, tracks candidate performance, and emails interview summaries

Advanced RAG architecture built using Transformers and TensorFlow to enabled credibility score checker for social media posts. Recent articles are web-scrapped from google-news-api using scrapy and stored in MongoDB, their titles are embedded in ChromaDB. This is accessed by LLM using Mistral API key.

RAGent Chatbot is a smart AI assistant that combines Retrieval-Augmented Generation (RAG) with a tool-using LLM agent. It can answer user queries using uploaded documents, perform web searches, summarize text, solve math expressions, check weather, and more — all powered by LangChain, Qdrant, and Gemini.

RAG code sorting search, RAG knowledge organization

There is documents searcher project based on Rust and Opensearch technologies.

A self-contained Google Colab Notebook that implements RAG using Postgres pgvector, pgai, and LLMs. Supports local-Ollama/OpenAI/Anthropic

A system that combines Retrieval-Augmented Generation (RAG), the Claude Sonet 3.5 LLM, and the Pathway framework to analyze financial reports and tables. It ingests data from Google Drive, processes both structured and unstructured formats in real time, and presents insights via a Streamlit-powered interface.

Snowflake-grounded natural-language chat agent over US Census data — guardrail → state resolver → hybrid metadata retrieval → structured plan + AST-validated SQL → self-check pipeline. Built as a 24-hour take-home in Next.js / TypeScript.

Llama_RAG_System is a local Retrieval-Augmented Generation (RAG) system that leverages the LLaMA model to provide intelligent answers to user queries by processing uploaded PDFs and fetching relevant web information while ensuring privacy.

A powerful platform designed to simplify the creation and management of Data Contracts, bridging systems for seamless data ingestion

🍔 Automates a food order from a json food menu using Retrieval-Augmented Generation (RAG) with OpenAI and MistralAI.

A chatbot that provides insights into the luxury fashion industry. It uses Cohere API and ChromaDB for efficient document retrieval and dynamic conversation.

C# SDK for the Tavily API -- AI-powered search and extraction

AI onboarding copilot with RAG, escalation levels, and MLflow LLMOps. FastAPI · Streamlit · LanceDB.

Think of ducky as your over-the-shoulder mentor. He watches your codebase and notifies you when you implement bad code. Ducky helps developers to understand code without handing them the complete answer.

Material y ejercicios del curso IMPACT#IA sobre LlamaIndex y Pydantic: desarrollo de agentes, uso de APIs, automatización y parseo de documentos con modelos estructurados.

Mindsage is an all offline journalling app

Python asynchronous SDK for managing RAGFlow APIs

ACL Findings 2025 | RaPID: retrieval-augmented long text generation with planning and information discovery.

HealthMate is an AI-powered voice agent that helps users get clear, reliable answers to health-related questions in real time. It combines FastAPI, AssemblyAI, and React with Retrieval-Augmented Generation (RAG) and VectorDB Cloud (Native) for fast and relevant responses.

A technical white paper on how LLMs handle memory, why context windows alone are not enough, and what production engineers need to know about memory architectures, security risks, and the road to in-weights personalisation.

A 100% offline, multi-agent AI swarm powered by Ollama. Features 5 specialized agents, hybrid RAG memory, and ChatDev integration. Protected by a strict 4-tier safety execution protocol to safely unleash uncensored Protocol 666 and offensive security Hat Mode algorithms.

Automates discovering and categorizing new Neovim plugins with RAG, creating PRs to yutkat/my-neovim-pluginlist.

OpenAI powered Reddit marketing tool for extracting latest Reddit comments from subreddits, sorted by "Hot", & pinpointing users who exhibit potential interest in your project or product.

Fault-tolerant multi-agent AI architecture: hierarchical agent coordination, edge-cloud hybrid inference, and adversarial chaos testing for systems that can't afford to fail.

Stop writing resumes from scratch. EvidenceCV turns your real work into an ATS-optimized resume using RAG. 🥉 3rd Place - eHacks Hackathon 2026.

OutcomeOps: Context Engineering for AI-Assisted Development

RAG挑战赛的获胜方案已修改为无需GPU即可运行。使用Mineru和Qwen API进行PDF解析和问答。非常适合在任何机器上学习RAG技术。

Code search and session observability for coding agents. Hybrid BM25+semantic, optional cross-encoder rerank, self-bench on your own data. npm install -g @sivru/cli

Effortlessly distill 20,000+ page legal and medical documents into concise, using Gemini 2.0 Flash Pro AI-powered summaries with our cutting-edge RAG system and NLP pipeline.

Semantic codebase understanding service — maps application structure, authentication patterns, authorization models, and data flows for intelligent security analysis.

local-first persistent memory for Claude Desktop, enabling private hybrid RAG search across past conversations without API calls

API service to search vectorized Cursor IDE chat history using LanceDB and Ollama

Search Swagger API Docs with LLM 🤖 (RAG-powered bot)

This repository features a CLIP-inspired contrastive model that aligns audio signals and transcripts in a shared embedding space, enabling bi-directional retrieval (Audio→Text and Text→Audio). It uses a frozen Whisper-large encoder for audio (with a lightweight trainable adapter) and a trainable Nomic-embed-text-v1.5 for text

The Image Search Engine is a modern AI-powered web application leveraging Retrieval-Augmented Generation (RAG) to combine large language models (LLMs) with advanced image retrieval techniques.

Retrieval-Augmented Generation server with Pinecone and OpenAI

Symfony RAG implementation featuring local embeddings with Ollama and pgvector. Inspired by the CodelyTV RAG course and built on the Symfony-Docker (FrankenPHP) skeleton using the neuron-ai library.

Persistent semantic memory for AI agents — SQLite + FTS5 + vector hybrid search, optional Qdrant

End-to-end GenAI framework with built-in evaluation, security, and modular productivity tooling.

WriteBee is an advanced Chrome extension that brings cutting-edge AI capabilities directly to your browser using Chrome's built-in AI APIs. Process text, analyze images, transcribe audio, chat with web pages and PDFs - all running locally on your device with complete privacy.

📈 AI-powered financial analyst with real-time stock data, SEC 10-K analysis, and LangGraph agents. Built with FastAPI, React, FAISS, and Yahoo Finance API.

Hybrid (FAISS + BM25) ensemble tool retriever with cross-encoder reranking for LLMs. Dynamically serves top-k tool schemas to save context tokens.

🚔 PoliceMitra: AI-Powered Law Enforcement Assistance System: PoliceMitra is a web-based system designed to support law enforcement agencies by integrating advanced AI technologies. It streamlines workflows, detects weapons in crime scene media, and retrieves legal information in real-time through a powerful RAG-based chatbot.

Chat with your documents. RAG implementation leveraging OpenAI's GPT-4o and Text-Embedding-3-Small models.

A prototype Healthcare Assistant using Retrieval-Augmented Generation (RAG) to provide primary health suggestions by retrieving data from a vector database or searching the internet when needed.

Match synthetic patients to clinicaltrials.gov restAPI data with ChatGPT4. Course project for Johns Hopkins University EN.705.651.8VL : Large Language Models: Theory and Practice

A Python project that deploys a Local RAG chatbot using Ollama API. Refines answers with internal RAG knowledge base, and uses both Embedding and LLM models.

In-depth demonstration of C# Semantic Kernel SQLiteVec Hybrid Search Tutorial - Audio Guide

Smart Context Optimization for LLMs - Reduce tokens by 66%, save 40% on API costs. Intelligent ranking and selection of relevant context using embeddings, keywords, and semantic analysis.

A FastAPI gateway for local LLMs that adds intelligent web research, multilingual recency/how-to detection, time-anchored guidance, context injection, and OpenAI-compatible SSE streaming. Turn any local model into a recency-aware, context-enhanced assistant instantly.

detoks는 LLM 사용 방식을 재설계하여 토큰과 컨텍스트를 최적화하는 CLI 시스템입니다.

Multi-disease segmentation chest X-rays by YOLO and DenseNet121, CoAtNet models

A comprehensive AI-powered chat system built with modern microservices architecture, featuring LangGraph, FastAPI, and a beautiful Vue.js frontend. The system provides enterprise-grade chat capabilities with advanced AI integration, user management, and administrative controls.

An AI Expert for your C# Codebase

Advance RAG is an enterprise-grade RAG system built on HippoRAG, designed for high-accuracy question answering in multilingual environments. It combines knowledge graph-based retrieval, hybrid search strategies, and advanced query processing to deliver precise, source-cited responses.

This is the Python backend for InsightAI, Architected a microservices-based EdTech platform combining Study Companion, Project Planner, and ShopGenie modules.

Battle-tested email archaeology application that allows you to search and compile reports from multiple accounts with advanced AI capabilities. NeuralMail does not let you compose or sent emails, it tackles the problem of finding things of which you know they "must be in some email".

Tensory — Claim-native MCP Memory Server. LLM claim extraction → Graph + Vector + SQLite. Zero-setup persistent memory for Claude, Cursor & any MCP client.

Self-hosted chess tutor that blends relational data (PostgreSQL) with vector search (Qdrant) to answer natural-language questions about annotated chess games. OCaml powers ingestion, hybrid retrieval, and CLI tooling.

듀듀의 포트폴리오 입니다